Customer Segmentation with K-Means Clustering¶

Id : #CC69853¶

Name : Utkarsh Upadhyay¶

Importing liberaries and modules¶

In [1]:
# Data
import numpy as np
import pandas as pd

# Data Visualization
import plotly.express as px
import matplotlib.pyplot as plt

# Data preprocessing
from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import StandardScaler

# Clustering Models
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA

Reading dataset¶

In [2]:
d = pd.read_csv('data.csv')
d
Out[2]:
CustomerID Gender Age Annual Income ($) Spending Score (1-100) Profession Work Experience Family Size
0 1 Male 19 15000 39 Healthcare 1 4
1 2 Male 21 35000 81 Engineer 3 3
2 3 Female 20 86000 6 Engineer 1 1
3 4 Female 23 59000 77 Lawyer 0 2
4 5 Female 31 38000 40 Entertainment 2 6
... ... ... ... ... ... ... ... ...
1995 1996 Female 71 184387 40 Artist 8 7
1996 1997 Female 91 73158 32 Doctor 7 7
1997 1998 Male 87 90961 14 Healthcare 9 2
1998 1999 Male 77 182109 4 Executive 7 2
1999 2000 Male 90 110610 52 Entertainment 5 2

2000 rows × 8 columns

Preprocessing¶

In [3]:
d.rename(columns={'Annual Income ($)': 'Income','Spending Score (1-100)':'Score'}, inplace=True)
In [4]:
d = d[['Gender', 'Age', 'Income','Score']]
In [5]:
d.describe()
Out[5]:
Age Income Score
count 2000.000000 2000.000000 2000.000000
mean 48.960000 110731.821500 50.962500
std 28.429747 45739.536688 27.934661
min 0.000000 0.000000 0.000000
25% 25.000000 74572.000000 28.000000
50% 48.000000 110045.000000 50.000000
75% 73.000000 149092.750000 75.000000
max 99.000000 189974.000000 100.000000
In [6]:
d = d[(d['Income'] > 15000) & (d['Age'] > 16) & (d['Score'] > 3)]
In [7]:
d.describe()
Out[7]:
Age Income Score
count 1596.000000 1596.000000 1596.000000
mean 56.921053 112080.638471 51.760025
std 24.141070 43764.532653 26.997632
min 17.000000 17000.000000 4.000000
25% 35.000000 76077.250000 30.000000
50% 57.000000 110045.000000 51.000000
75% 78.000000 149628.000000 75.000000
max 99.000000 189974.000000 100.000000
In [8]:
d.shape
Out[8]:
(1596, 4)

Graphical Representation¶

Male and Female Ratio¶

In [9]:
import plotly.express as px
px.pie(values=d['Gender'].value_counts(), names=d['Gender'].value_counts().index)

Income Range¶

In [10]:
px.histogram(d, x='Income', marginal='violin')

Purchase score range¶

In [11]:
px.histogram(d, x='Score', marginal='violin')

Age range¶

In [12]:
px.histogram(d, x='Age', marginal='violin')

Updating dataframe columns¶

In [14]:
d = d[['Age', 'Income','Score']]
In [15]:
scaler = StandardScaler()
full_data = scaler.fit_transform(d)
In [16]:
full_data
Out[16]:
array([[-1.48843078, -1.76181015,  1.08339647],
       [-1.52986694, -0.59611771, -1.69549562],
       [-1.40555844, -1.213249  ,  0.93518889],
       ...,
       [ 1.24635632, -0.48272555, -1.39908047],
       [ 0.83199464,  1.60061827, -1.76959941],
       [ 1.37066482, -0.03361396,  0.00889153]])

Principal Component Analysis¶

In [18]:
pca = PCA(n_components=2, random_state=42)

# Fit and transform the data to obtain the 2D projection
data = pca.fit_transform(full_data)

Elbow Method : determining the optimal number of clusters¶

In [19]:
ssd = []

# fit KMeans clustering with different values of k
for k in range(1, 11):
    kmeans = KMeans(n_clusters=k, random_state=42)
    kmeans.fit(full_data)
    ssd.append(kmeans.inertia_)

# create a dataframe with the k values and corresponding ssd
d = pd.DataFrame({'k': range(1, 11), 'ssd': ssd})

# create the line plot using Plotly Express
fig = px.line(d, x='k', y='ssd', title='Elbow Method')
fig.update_traces(mode='markers+lines', marker=dict(size=8))
fig.show()

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.
In [20]:
kmeans = KMeans(n_clusters=2, random_state=42)

kmeans.fit(full_data)

labels = kmeans.labels_
centroids = kmeans.cluster_centers_

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning:

The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning

C:\Users\utkar\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:1382: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=7.

K-Means clustering visualization¶

In [22]:
fig = px.scatter(
    x=data[:, 0], y=data[:, 1],
    color=labels,
    size_max=5, 
    opacity=0.7,
    labels={'x':'X', 'y':'Y'},
    color_continuous_scale=['black', 'cyan'])

# Add a trace for the cluster centers
import plotly.graph_objs as go

fig.add_trace(
    go.Scatter(
        x=centroids[:,0],
        y=centroids[:,1],
        mode='markers+text',
        text=['Centroid 1', 'Centroid 2'],
        marker=dict(
            size=20,
            color='orange',
            opacity=1.0,
            symbol='diamond'
        )
    )
)

# Update the layout
fig.update_layout(
    coloraxis_showscale=False,
    title='K Means Clustering Visualization'
)

# Show the plot
fig.show()